LiBSU - Redesigning the "Battery Support Unit" to Support Lithium

Background:
1: I apologize in advance for any incorrect vernacular specific to the Prius. I'm a (passionate) G1 Honda Insight owner, who prior to today had never been inside a Gen3 Prius. Prior to last night I'd never even studied the wiring diagram... so I'm bound to get some terms wrong.

2: I designed and sell a replacement computer that allows the G1 Honda Insight to safely use modern lithium traction batteries. My long term vision is to eventually offer a similar computer for the various NiMH Prius models, starting with Gen3. For now I'm going to call it 'LiBSU' because it replaces the OEM "Battery Smart Unit".

3: Note that I'm well aware that NexCell exists; theoretically the (unreleased) LiBSU hardware is actually compatible with those lithium modules (e.g. by removing the PCB inside each NexCell module, and then connecting a wire harness from LiBSU to each NexCell wire harness). However, I don't intend to commercialize that solution, as NMC lithium has much higher energy density than LFP.

4: LiBSU isn't my top priority right now... it's a parallel side passion to an overgrown hobby project. I'm essentially retired, although I'll never stop tinkering. Expect infrequent updates over the next several months. For example, I don't intend to gather anymore data for a few weeks, as I'm leaving for vacation next week. Eventually LiBSU will exist "when it's ready", but until then it won't.

...

Below are my first impressions regarding adding lithium to the Gen3 Prius hybrid system:
A: I should only need to replace the "Battery Support Unit", which is surprisingly simple... all it does is monitor cell voltages, battery current, four temperature sensors, and test for HVDC isolation faults.
B: The BSU has a differential (RS-485?) serial bus. Looks like there are QTY10 frames per second, with nothing too difficult to decode (of course this will take time). On the PHY, we have an A-B diff pair operating between 1.5 (low) and 3.5 (high) volts.
C: Looks like we'll need QTY60 NMC lithium cells to approximate the OEM QTY168 NiMH cell voltage range.
D: The BSU is much smaller than I thought it would be... cramming QTY60 ADC channels into the OEM steel enclosure might not be possible, but there's certainly plenty of additional room with a replacement outer enclosure. Space is still going to be tight, but I know it's possible.
E: For now, I intend to place all BMS circuitry on the LiBSU PCB, rather than on distributed PCBs attached to each module. This makes it easier to switch between different batteries... of course, it also increases the replacement BMS wire harness cost. I'm willing to use distributed PCBs if we can find a large source of "fits-like-a-glove" lithium modules from another car. I have a few leads on this, but not worth thinking about now; designing LiBSU is agnostic to the actual lithium cell/chemistry used.
F: I was surprised to deduce that the OEM BMS wire harness uses RED/RED wire for HVDC-, and BLK/BLK wire for HVDC+... not the first time I've seen this, but ok. Also was surprised that the service disconnect plug doesn't cut the pack in half... QTY4 NiMH pairs on the bottom end and QTY10 on the top end.
G: On the BSU's 8p 'serial' connector, I wasn't able to determine what the 2.66 VDC signal was used for. It certainly had an 'intentional' voltage ramp at keyOn, but otherwise it was always 2.66 VDC.
H: From a hardware perspective, deriving LiBSU from my existing LiBCM design is trivially easy. In fact, I should be able to prototype a functional prototype with an LiBCM PCB. It won't fit mechanically, but I'm sure I won't be the first person to drive a Prius without the rear seats installed .
J: I'll certainly be able to use LiBCM to 'sniff' boatloads of serial data from OEM BSU. Decoding this data will certainly take time, but nothing jumped out as 'impossible' during my cursory inspection. Notably, I have all the information required to design the RevA PCB.

Overall, I don't anticipate any major headaches deriving LiBSU from my existing LiBCM project.

...

So what is LiBSU?
-A drop-in replacement computer that allows a Gen3 NiMH Prius to safely use a lithium traction battery, and;
-built-in level 1 (~1500 watts) PHEV charger (copied directly from existing LiBCM design), and;
-built-in advanced diagnostics via USB. Of course, anything Dr.Prius can presently access will continue to work, too... but you'll need to use USB to read all QTY60 cell voltages (because the OEM system only supports QTY14 channels).

Thanks for reading. Please let me know if I got anything wrong. Y'all are the domain experts. I'm open to any and all LiBSU-related conversations. Grab your popcorn and let's discuss.

 

Gen2 is on the roadmap, after Gen3. The modifications are trivial; primarily the OEM serial bus PHY is different, but otherwise the architecture is the same. Gen2 should ship shortly after Gen3... but Gen3 is going to take a while, as I outlined in my original post.

 

Thanks for your support!

Hopefully your existing traction battery can hang in there for at least several more months...
...even though I'm deriving LiBSU from an existing project (LiBCM), properly engineered products take time.

 

Minor update:
Today I deciphered the BSU's differential serial frame structure. Decoding the frame order is going to take some time, but I should be able to man-in-the-middle the existing data with my prototype hardware fairly soon. I should also be able to do some rudimentary deciphering on the bench... I'll probably have the "happy path" worked out before I actually test in the car (I still don't own one, but a fellow Chattanoogan I met does). I'll then need to artificially induce numerous hardware faults to figure out how to properly transmit "something is borked" signals.

After I decode the serial bus, the rest is pretty straight forward:
-the actual BMS electronics are essentially copy-pasted from my existing Honda Insight design, and;
-I need to select a suitable donor lithium pack, but that's ultimately a function of mechanical fitment. Already have a few ideas on which modules I'll recommend, but no public comment until I'm 100% certain (lest someone purchase a pack that I don't ultimately use), and;
-eventually I'll need to redesign my existing LiBCM circuit board so that it fits inside the OEM BSU enclosure. I looked inside a Gen2 Prius pack last week and was surprised how much larger the BSU is versus Gen3.

As I wrote in my initial post, I don't presently have any timeline regarding when LiBSU is available for purchase... certainly not any time soon, as this isn't my top priority right now. This is a passion project, plus I'm retired (but will never stop working).

 

Thanks for the additional Toyota-specific info.
In general, I'm designing the LiBSU schematic to be compatible with Toyota's entire hybrid fleet... obviously the connectors and mechanicals will change between vehicles (and lithium modules), but my goal is to design a Swiss Army Knife.

Last night I further deciphered the Gen3 Prius' serial bus. I have it mostly framed out at this point... not that complicated overall, but obviously there are lots of corner cases (and failure modes) to induce, so that LiBSU can safely replace the safety critical OEM computer. I will of course do as much testing as I can on the bench, since gathering data is much more difficult in the car.

 

Thanks for the information regarding past attempts. Reading over the wiki you linked really makes me wish that PHEV's had won out over pure electric vehicles. I may be new to the Prius community, but I can't wait to contribute to keeping these aging NiMH vehicles on the road.

Besides the two youtube videos, I wasn't able to find much info on ewertenergy. Their design predates mine by fifteen years, so it makes sense that I've never heard of them. If anyone from Ewert Energy is still around on priuschat I'd love to say hello.

 

Another small update:
I've finished mapping out all the cell voltages, temperature sensors, and current sensor values in the serial data stream. Still deciphering Toyota's particular checksum routine... @turnip73's latest post had a few leads on that front, but I haven't gotten around to testing them out yet. Just a minor hurdle in the process.

I'll note that Toyota's data packing method in really wonky. They send some blade voltages more often than others; some are double-stuffed into the same 'major' byte, while others are all on their own. One blade voltage is actually sent incorrectly (or at least it doesn't follow the same general format; specifically, it has an extra 'minor' byte). I assume the receiver's firmware is programmed to fix that issue... not sure why they didn't fix it on the BSU firmware, but I'll make sure to carry it through to LiBSU.

They could have reliably sent the same information using 3x fewer bytes, but whatever... that's what they chose to do. I still haven't tested in the car, so while I know which bytes represent error codes, I don't actually know what each specific error code is yet. All in due time.